Knowing today what will work tomorrow
Two data experts reveal how expertise can help thousands of people lead better lives - for example, by developing a dashboard that already knows today which drugs will cure tomorrow
Stefanie Kaufmann studied bioinformatics in Munich with a Bachelor's and a Master's degree and subsequently completed her doctorate in bioinformatics at the TUM. After completing her doctorate, she worked for two years as a management consultant for Detecon in the field of energy. She joined Roche in January 2017 as a Data Scientist.
Anna Bauer-Mehren studied Bioinformatics (Diploma/Master) at LMU and TUM. She then earned her doctorate in biomedical informatics at the University of Pompeu Fabra in Barcelona and worked for two years as a postdoctoral fellow at the Stanford School of Medicine. After several years abroad, Anna moved to Munich in 2013, where she has headed the Data Science department since 2014.
At Data Science, you don't first think of pharmaceutical or biotechnology companies. Why does Roche need data specialists?
Stefanie Kaufmann: Roche's declared goal is to offer patients an individualized therapy - and for this we need data. Because every disease and every patient is different: It is said that a person has "cancer", but not all cancers are the same! There are many different subtypes of cancer that need to be treated differently. "Personalized therapy" is the word of the day - and this is where data comes into play and people who can handle it.
Anna Bauer-Mehren: The more accurate our data, the better we can treat: That's the idea behind personalized medicine, and that's why data experts are so important to us today. With the Companion Diagnostics concept, for example, our data experts have for years been able to find out which patient groups respond to a particular Roche drug. Therapy and diagnostics work hand in hand. It is precisely this combination that makes Roche so special.
What exactly do you mean when you talk about data at Roche?
Stefanie Kaufmann: On the one hand, of course, data on patients, disease histories and the disease in question, e.g. biomarkers, laboratory values and histopathological images. Data from wearables also play an important role in the early detection of certain diseases. On the other hand, data from preclinical development, research and basic research are central. It is in this interaction that the individualised medicine that Roche is working towards emerges.
Anna Bauer-Mehren: At the moment and against the background of the best possible individualisation, one of our goals is to collect as much information as possible about a patient. Using sensors such as smartphones or fitness wristbands, for example, we collect data on the movement and speech of Parkinson's patients and healthy people. Using this data, we learn, for example, how a Parkinson's patient differs from another Parkinson's patient or from a healthy person and how our medication works. In the digital world, we can continuously collect data and thus get much closer to reality.
How do you and your department contribute to the goal of individualised therapies?
Anna Bauer-Mehren: I am head of the Data Science department at pREDi (Pharma Research and Early Development Informatics), which is Roche's Pharma Research Informatics department. Our department oversees all steps from data collection to evaluation and supports all Roche research areas, from neuroscience and oncology to our biology and chemistry laboratories that produce the drugs. One of our subdivisions ensures that the data is collected and output correctly. Other colleagues take care of the data flow and provide the right system landscape, software and infrastructure to collect data in a structured manner, make it anonymous and guarantee its quality. They implement software that already exists on the market, but also develop their own solutions. My Data Science department is responsible for evaluating the data. We want to understand and interpret the data and, for example, find a biomarker that explains why some patients respond to a drug and others do not.
Stefanie Kaufmann: My department, Scientific Information Services, is an internal service department that advises customers from both Roche Pharmaceuticals and Roche Diagnostics. Our main task is to support scientists at Roche in managing the flood of information and data in the biomedical field. Every day about 2,400 bio-medical publications are published, along with hundreds of experiments and the corresponding data. We help scientists use digital methods to extract what is relevant to them. The focus is on text and data mining, but also on machine learning and the automated evaluation of publications.
Who is Roche looking for to achieve its goals in data-based medicine?
Anna Bauer-Mehren: As a healthcare company, Roche is generally looking for applicants with a background in information technology and life sciences, for example in a minor subject or a course of study such as biotechnology or bioinformatics. This is important because Roche collaborates with doctors, biologists and chemists in all its projects, whose language and processes should be understood. Experts in big data, data analysis, machine learning, robotics, automation and digitisation are also becoming increasingly important. When it comes to graduation, we often hire people after the master's degree, but more often after the Ph.D. or postdoc.
Stefanie Kaufmann: We also rely on domain knowledge on the one hand and experience with data science on the other. Many courses prepare students for this, for example bioinformatics, biostatistics or biophysics. However, we also employ computer scientists who have built up domain knowledge through a minor or a dual course of study. Physicists also often have an affinity for biology. In any case, we also require a certain level of biomedical expertise so that our employees can discuss with customers at eye level. Although a doctorate is not a must, it is an advantage when starting out in a research company like Roche.
Employers all over the world are looking for experts in data science, statistics, mathematics and IT. Why Penzberg and not Silicon Valley, for example?
Anna Bauer-Mehren: Despite many offers, I left Silicon Valley in favour of Penzberg. I wanted to work close to medicine and see what our work does for patients and what it improves. Of all the pharmaceutical and biotechnology companies, I chose Roche because of its global leadership in oncology, among other things. The Penzberg site, one of Europe's largest biotech sites, was also important: it combines pharmaceutical and diagnostics research with production. The Munich area is also interesting, especially the universities with which we work - an important argument for all those who want to work close to research.
Stefanie Kaufmann: What sets us apart the most from the Googles and Facebooks of the world in my eyes is our goal. We are a research company with the goal of making a valuable contribution to society and improving the quality of life of many patients with innovative drugs and diagnostic procedures. Working in Silicon Valley is certainly fun, but it's a different feeling when your work contributes to improving and prolonging life in the long run. My second reason would be that in Penzberg we combine many business sectors at one location: Pharmaceuticals and Diagnostics, Development and Production. For Data Scientists, this results in an extremely wide range of tasks.
04.07.2018 / © Roche with e-fellows.net